Quantification of patient motion in FDG-PET/CT brain imaging

Abstract

Objectives FDG-PET is an important imaging tool in evaluation of neurological disorders, including drug-resistant epilepsy. Head motion can degrade PET image quality; however, the extent in routine examinations remains insufficiently characterized. This retrospective study aimed to quantify the magnitude and pattern of head motion in brain FDG-PET/CT and establish displacement-based scores for describing motion severity. Materials & Methods The analysis included 140 patients with focal drug-resistant epilepsy undergoing brain FDG-PET/CT, with 27 examined under general anesthesia as a reference group. PET acquisition comprised five consecutive 2-min frames. Head motion was quantified by rigid PET-to-PET registration and expressed as maximal displacement. Two experienced readers independently graded motion on a four-point ordinal scale. Ordinal logistic regression with repeated 10-fold cross-validation was used to derive displacement thresholds corresponding to visual motion scores. Results Patients without anesthesia demonstrated greater maximal displacement than the reference group (median, 2.6 vs. 0.7 mm; p < 0.001). Displacement increased throughout acquisition, peaking in the final frame in 85.7% of examinations. Motion predominantly involved inferior-superior translation and rotation around left-right axis, corresponding to head flexion-extension. Maximal displacement was strongly associated with visual motion grades (Kendall's Tau;=0.68-0.70; p < 0.001). Derived thresholds classified displacement >3.25 mm as moderate and >6.4 mm as severe motion. In total, 40.7% of patients without anesthesia demonstrated moderate or severe motion, with severe motion in 12.4%. Conclusion Clinically relevant head motion was frequent in brain FDG-PET/CT despite head fixation, predominantly characterized by flexion-extension. Registration-based displacement provided an objective measure closely associated with expert visual assessment.